A hybrid framework combining background subtraction and deep neural networks for rapid person detection

Kim, Chulyeon; Lee, Ji-Young; Han, Taekjin; Kim, Young Min

doi:10.1186/s40537-018-0131-x

Cited by 63 publications

(40 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Kim et al [13] proponen un sistema para la detección rápida de personas mediante la técnica de sustracción de fondo, que permite optimizar el costo computacional en comparación a circuitos cerrados de vigilancia en las grandes urbes, obteniendo rendimientos superiores al 82% debido a la cantidad de falsos positivos al momento de la clasificación final. Asimismo, Jeon et al [14] implementan un sistema de sustracción de fondo que, mediante filtrados de niveles de gris, diferenciación de pixeles, acople según el brillo de la imagen y umbralización para las detecciones, permite obtener un porcentaje aproximado de acierto en las de-π C. V. Niño, S. A. Castro, B. Medina & D. Guevara tecciones de 93%.…”

Section: Metodologíaunclassified

Análisis de viabilidad y diseño de un sistema electrónico para el seguimiento de la dinámica poblacional en la ciudad de Cúcuta

Rondón¹,

Casadiego²,

Delgado³

et al. 2022

Ing.USBMed

View full text Add to dashboard Cite

En este documento se muestran los resultados obtenidos al realizar un análisis de viabilidad y el diseño de un sistema electrónico para el seguimiento de la dinámica poblacional en la ciudad de Cúcuta. El análisis de viabilidad se realizó al indagar respecto a la posición de la ciudadanía en lo que compete al comportamiento de la dinámica poblacional, su impacto en la región y la preferencia por el uso de la tecnología como solución de problemas sociales. Asimismo, el diseño del sistema electrónico de procesamiento de señales se realizó seleccionando las herramientas de hardware y software por medio de una matriz de priorización, comparando once factores de interés, que permitieron elaborar los algoritmos de detección de personas basado en las técnicas de Histograma de Gradientes Orientados y de Sustracción de Fondo con Extracción de Blobs, posteriormente probados con imágenes propias de los autores.

show abstract

Section: Metodologíaunclassified

Análisis de viabilidad y diseño de un sistema electrónico para el seguimiento de la dinámica poblacional en la ciudad de Cúcuta

Rondón¹,

Casadiego²,

Delgado³

et al. 2022

Ing.USBMed

View full text Add to dashboard Cite

show abstract

“…However, they remain too slow for real-time operation on embedded devices [4]. Some implementations that attempt to combine deep-learning approaches with Background Subtraction (BS) [1]. Despite this, they require a GPU module for real-time detection.…”

Section: Related Work and State Of The Artmentioning

confidence: 99%

“…• an object intersects a ground surface in real world in at least one point, and To compensate for the camera lens distortion, the correction procedure is performed as a first step. The pixels of interest are corrected taking into account radial and tangential lens distortion factors using Equation (1). Each pixel has coordinates u, v in image plane along horizontal and vertical axis correspondingly.…”

Section: Object Distance Obtainingmentioning

confidence: 99%

“…Some of these sensors are camera-based and used in services of a smart city, such as surveillance or traffic monitoring. The cameras produce a lot of data which is processed centrally by powerful computers that are equipped with GPUs [1]. The services using camera as a sensor might need to process data decentralized, locally on IoT devices, which is primarily caused by the implementation of the Internet connection.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Fast Object Detection Using Dimensional Based Features for Public Street Environments

et al. 2020

View full text Add to dashboard Cite

Modern object recognition algorithms have very high precision. At the same time, they require high computational power. Thus, widely used low-power IoT devices, which gather a substantial amount of data, cannot directly apply the corresponding machine learning algorithms to process it due to the lack of local computational resources. A method for fast detection and classification of moving objects for low-power single-board computers is shown in this paper. The developed algorithm uses geometric parameters of an object as well as scene-related parameters as features for classification. The extraction and classification of these features is a relatively simple process which can be executed by low-power IoT devices. The algorithm aims to recognize the most common objects in the street environment, e.g., pedestrians, cyclists, and cars. The algorithm can be applied in the dark environment by processing images from a near-infrared camera. The method has been tested on both synthetic virtual scenes and real-world data. The research showed that a low-performance computing system, such as a Raspberry Pi 3, is able to classify objects with acceptable frame rate and accuracy.

show abstract

“…But, false-positive rate of detection was not condensed. A simple framework was presented in [5] to identify moving objects by fusing backdrop subtraction and Convolutional Neural Networks (CNNs). However, dynamic background during detection was not considered.…”

Section: Introductionmentioning

confidence: 99%

Anisotropic Sophisticated Spatio-Temporal Contours Based Deep Neural Learned Moving Objects Detection in Video

Kader*,

Nihal

2019

IJRTE

View full text Add to dashboard Cite

Object detection in the video sequence is a significant problem to be resolved in image processing because it used different applications in video compression, video surveillance, robot technology, etc. Few research works have been designed in conventional works to discover moving objects using various machine learning techniques. However, dynamic changing background, object size variations and degradation of video frames during the object detection process remained an open issue. In order to overcome such limitations, Anisotropic Sophisticated Spatiotemporal Contours based Deep Neural Network Learning (ASSC-DNNL) practice is projected. ASSC-DNL Technique initially obtains a number of video file as input at the input layer. After acquiring the video, input layer forward it to hidden layers. Subsequently, ASSC-DNL Technique accomplishes the encoding process in the first hidden layer using Anisotropic Stacked Autoencoder (ASA). During the encoding process, ASSC-DNL practice maps each video frames pixels in input video via code. This practice results in compressed video with enhanced quality. Afterward, ASSC-DNL practice transforms compressed video into a numeral of frames in the second concealed layer. Followed by, ASSC-DNL practice carried out Teknomo–Fernandez Spatiotemporal Based Background Subtraction (TS-BS) process at the third hidden layer, in which it effectively segments the foreground images from dynamic changing background. Then, ASSC-DNL practice deep analyzes the foreground image of video frames and mines some features like shape, color, texture, intensity, and size. Finally, ASSC-DNL Technique exactly finds the moving objects in video frames according to identified features with minimal time at the output layer. Therefore, ASSC-DNL Technique obtains enhanced moving objects detection performance when compared to existing works. The simulation of ASSC-DNL practice is conducted via different metrics such as accuracy, time and false positive rate towards in detection.

show abstract

A hybrid framework combining background subtraction and deep neural networks for rapid person detection

Cited by 63 publications

References 61 publications

Análisis de viabilidad y diseño de un sistema electrónico para el seguimiento de la dinámica poblacional en la ciudad de Cúcuta

Análisis de viabilidad y diseño de un sistema electrónico para el seguimiento de la dinámica poblacional en la ciudad de Cúcuta

Fast Object Detection Using Dimensional Based Features for Public Street Environments

Anisotropic Sophisticated Spatio-Temporal Contours Based Deep Neural Learned Moving Objects Detection in Video

Contact Info

Product

Resources

About